Set up

Import Data

Tables by Severity

Demographic

The table below provides demographic information based on severity of injury

# level naming for categorical variables
df_demo$gender <- factor(df_demo$gender,
                   levels = c(1,2,3),
                   labels = c("Male", "Female", "Nonbinary"))

df_demo$work_current <- factor(df_demo$work_current,
                   levels = c(1,0),
                   labels = c("Yes", "No"))

df_demo$severity <- factor(df_demo$severity,
                   levels = c(2,3),
                   labels = c("Moderate", "Severe"))

df_demo$mech_injury <- factor(df_demo$mech_injury,
                   levels = c(1,2,3,4,5),
                   labels = c("Fall", "MVC", "Sports", "Violence", "Pedestrian struck"))

df_demo$income <- factor(df_demo$income,
                   levels = c(1,2,3),
                   labels = c("<52K", "52K-156K", ">156K"))

df_demo$marital_status <- factor(df_demo$marital_status,
                                 levels = c(1, 2, 3, 4),
                                 labels = c("Single", "Married", "Divorced", "Widowed"))
Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
Age (years) 46 51 (14) 44 (14) 0.091
Time since TBI (years) 46 7 (5) 10 (8) 0.14
Gender 46

0.20
    Male
7 (37%) 16 (59%)
    Female
11 (58%) 11 (41%)
    Nonbinary
1 (5.3%) 0 (0%)
Education (years) 46 15.47 (2.04) 15.11 (2.61) 0.60
Race/Ethnicity 46

0.24
    Asian
0 (0%) 2 (7.4%)
    Biracial
2 (11%) 0 (0%)
    Black
0 (0%) 1 (3.7%)
    Hispanic
1 (5.3%) 3 (11%)
    White
16 (84%) 21 (78%)
Employment status 46

0.69
    Yes
9 (47%) 10 (37%)
    No
10 (53%) 17 (63%)
Annual household income 46

0.84
    <52K
6 (32%) 10 (37%)
    52K-156K
9 (47%) 13 (48%)
    >156K
4 (21%) 4 (15%)
Size household 46 2.00 (1.05) 2.19 (1.36) 0.61
Marital status 46

0.11
    Single
5 (26%) 15 (56%)
    Married
11 (58%) 8 (30%)
    Divorced
3 (16%) 4 (15%)
    Widowed
0 (0%) 0 (0%)
Substance use score 46 4.16 (3.62) 1.67 (1.92) 0.011
Cause of injury 46

0.12
    Fall
10 (53%) 5 (19%)
    MVC
4 (21%) 11 (41%)
    Sports
1 (5.3%) 4 (15%)
    Violence
1 (5.3%) 4 (15%)
    Pedestrian struck
3 (16%) 3 (11%)
1 Mean (SD); n (%)
2 Welch Two Sample t-test; Pearson’s Chi-squared test

ACS

ACS3 (activity re-engagement scores - outcome measure) by severity of injury

Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
ACS Global Before 46 72 (11) 68 (10) 0.18
ACS Global Current 46 54 (17) 52 (12) 0.63
Global Retained (%) 46 75 (19) 77 (15) 0.72
ACS IADL Before 46 22.16 (2.27) 20.96 (3.29) 0.15
ACS IADL Current 46 17.9 (4.9) 16.8 (4.4) 0.43
IADL Retained (%) 46 81 (19) 81 (18) 0.98
ACS Leisure Before 46 22.5 (5.8) 20.9 (4.6) 0.33
ACS Leisure Current 46 18.0 (6.0) 16.6 (5.0) 0.41
Leisure Retained (%) 46 82 (22) 80 (16) 0.75
ACS Fitness Before 46 13.2 (4.4) 12.8 (4.4) 0.76
ACS Fitness Current 46 8.3 (4.7) 8.4 (3.3) 0.93
Fitness Retained (%) 46 64 (32) 69 (34) 0.55
ACS Social Before 46 13.95 (1.22) 12.89 (1.55) 0.013
ACS Social Current 46 9.68 (3.08) 9.91 (2.48) 0.79
Social Retained (%) 46 69 (19) 77 (18) 0.16
1 Mean (SD)
2 Welch Two Sample t-test

Below is the ttest for the specific t and p value for the difference between ACS3 previous social score, which was significantly different.

## 
##  Welch Two Sample t-test
## 
## data:  df_mod$acss_prev and df_severe$acss_prev
## t = 2.5819, df = 43.36, p-value = 0.01328
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.231927 1.885032
## sample estimates:
## mean of x mean of y 
##  13.94737  12.88889

FrSBe

Comparison of self-regulation scores by severity

Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
Executive function 46 41 (10) 42 (11) 0.82
Disinhibition 46 32.3 (6.1) 33.0 (5.9) 0.67
Apathy 46 33 (8) 34 (9) 0.61
Total FrSBe Score 46 106 (19) 109 (21) 0.65
1 Mean (SD)
2 Welch Two Sample t-test

TBI QOL

Comparison of subscales of TBI QOL measure by severity of injury

Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
Participation SRA 46 47 (7) 46 (6) 0.48
Anger 46 53 (9) 50 (10) 0.29
Anxiety 46 58 (7) 54 (10) 0.13
Communication 46 46 (9) 46 (10) 0.95
Depression 46 55 (8) 53 (11) 0.57
Dyscontrol 46 52 (8) 50 (9) 0.44
EF 46 34.9 (4.8) 35.9 (6.9) 0.56
Fatigue 46 57 (8) 54 (9) 0.29
Gen Cognition 46 36 (8) 37 (10) 0.74
Headache 46 51 (9) 48 (9) 0.41
Mobility 46 47 (10) 44 (8) 0.28
Pain 46 58 (11) 54 (10) 0.21
Positive Effect 46 50 (6) 50 (8) 0.99
Resilience 46 49 (6) 48 (9) 0.56
Satisfaction SRA 46 46 (6) 45 (7) 0.49
Self esteem 46 47 (11) 49 (11) 0.69
Stigma 45 50 (8) 51 (7) 0.62
Upper Extremity 46 47 (9) 42 (8) 0.078
1 Mean (SD)
2 Welch Two Sample t-test

TBI Composite

Comparison of composite scores for TBI QOL by severity of injury. Composite scores were calculated using:

Tyner, C. E., Boulton, A. J., Sherer, M., Kisala, P. A., Glutting, J. J., & Tulsky, D. S. (2020). Development of Composite Scores for the TBI-QOL. Arch Phys Med Rehabil, 101(1), 43-53. https://doi.org/10.1016/j.apmr.2018.05.036

Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
Physical Health Index 46 91 (14) 97 (14) 0.19
Emotional Health Index 46 97 (12) 101 (15) 0.26
Cognitive Health Index 46 93 (13) 95 (16) 0.74
Social Health Index 46 94 (12) 91 (13) 0.41
Global Health Index 46 93 (13) 95 (14) 0.52
1 Mean (SD)
2 Welch Two Sample t-test

Predictive variables

Table 2 in dissertation

This table compares only the Personal and Environmental Protective factors and self-regulation outlined in the dissertation. Note that the Cognitive Health Composite score was not used as it includes executive functioning, which in this paper is considered a self-regulatory process. Therefore, general cognitive functioning was used which assesses memory and concentration.

Characteristic N Moderate, N = 191 Severe, N = 271 p-value2
Physical Health Index 46 91 (14) 97 (14) 0.19
Emotional Health Index 46 97 (12) 101 (15) 0.26
General Cognition 46 36 (8) 37 (10) 0.74
Extraversion 46 7.16 (2.50) 6.78 (2.29) 0.60
Agreeable 46 7.11 (1.94) 7.15 (2.11) 0.94
Consciousness 46 8.16 (1.54) 7.67 (1.92) 0.34
Neuroticism 46 6.47 (2.20) 6.15 (2.66) 0.65
Openness 46 8.53 (2.09) 7.37 (1.86) 0.062
Annual household income 46

0.84
    <52K
6 (32%) 10 (37%)
    52K-156K
9 (47%) 13 (48%)
    >156K
4 (21%) 4 (15%)
Marital status 46

0.11
    Single
5 (26%) 15 (56%)
    Married
11 (58%) 8 (30%)
    Divorced
3 (16%) 4 (15%)
    Widowed
0 (0%) 0 (0%)
Social Support 46 84 (10) 76 (11) 0.019
Executive function 46 41 (10) 42 (11) 0.82
Disinhibition 46 32.3 (6.1) 33.0 (5.9) 0.67
Apathy 46 33 (8) 34 (9) 0.61
Total score 46 106 (19) 109 (21) 0.65
1 Mean (SD); n (%)
2 Welch Two Sample t-test; Pearson’s Chi-squared test

Below is the t-test for SPS total, which was significantly different between severity of injury

## 
##  Welch Two Sample t-test
## 
## data:  df_mod$spstotal and df_severe$spstotal
## t = 2.4526, df = 41.168, p-value = 0.01851
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##   1.368493 14.124685
## sample estimates:
## mean of x mean of y 
##  83.89474  76.14815

Descriptive Statistics

Descriptive statistics for each variable of interest for the data set including mean, median, SD, and IQR, kurtosis and se

vars n mean sd median trimmed mad min max range skew kurtosis se
record_id* 1 46 23.50 13.42 23.50 23.50 17.05 1.0 46.0 45.0 0.00 -1.28 1.98
age_current 2 46 47.07 14.44 46.50 46.97 18.53 21.0 72.0 51.0 0.17 -1.18 2.13
age_injury 3 46 38.04 14.95 34.50 37.32 17.05 18.0 66.0 48.0 0.36 -1.27 2.20
time_injury 4 46 9.09 7.10 7.00 8.20 5.93 1.0 30.0 29.0 1.14 0.73 1.05
gender 5 46 1.52 0.55 1.50 1.50 0.74 1.0 3.0 2.0 0.31 -1.15 0.08
race* 6 46 4.57 1.05 5.00 4.84 0.00 1.0 5.0 4.0 -2.39 4.56 0.15
edu 7 46 15.26 2.37 16.00 15.29 2.97 10.0 20.0 10.0 -0.19 -0.68 0.35
work_current 8 46 0.41 0.50 0.00 0.39 0.00 0.0 1.0 1.0 0.34 -1.92 0.07
hours_work 9 19 30.50 13.04 40.00 31.50 0.00 4.0 40.0 36.0 -0.74 -1.19 2.99
occ_years_pos 10 18 7.43 9.21 3.50 6.47 5.04 0.1 30.0 29.9 1.47 1.14 2.17
diff_occ 11 19 0.53 0.51 1.00 0.53 0.00 0.0 1.0 1.0 -0.10 -2.09 0.12
no_occ_stat 12 27 3.19 0.74 3.00 3.13 0.00 2.0 5.0 3.0 1.40 1.67 0.14
income 13 46 1.83 0.71 2.00 1.79 1.48 1.0 3.0 2.0 0.25 -1.05 0.10
house_size 14 46 2.11 1.23 2.00 1.92 1.48 1.0 7.0 6.0 1.68 3.66 0.18
children 15 46 0.43 0.50 0.00 0.42 0.00 0.0 1.0 1.0 0.25 -1.98 0.07
num_child 16 20 2.00 1.26 2.00 1.75 1.48 1.0 5.0 4.0 1.21 0.49 0.28
severity 17 46 2.59 0.50 3.00 2.61 0.00 2.0 3.0 1.0 -0.34 -1.92 0.07
mech_injury 18 46 2.39 1.39 2.00 2.26 1.48 1.0 5.0 4.0 0.71 -0.85 0.20
mech_injury_other* 19 6 1.17 0.41 1.00 1.17 0.00 1.0 2.0 1.0 1.36 -0.08 0.17
substance 20 46 2.70 2.99 2.00 2.24 2.97 0.0 13.0 13.0 1.33 1.75 0.44
acsg_prev 21 46 69.33 10.20 68.00 69.11 10.38 48.0 93.0 45.0 0.28 -0.61 1.50
acsg_curr 22 46 52.59 14.19 49.25 51.74 13.94 29.5 85.5 56.0 0.54 -0.42 2.09
acsg_retain 23 46 75.98 16.47 74.50 76.18 14.08 43.0 107.0 64.0 -0.09 -0.45 2.43
acsi_prev 24 46 21.46 2.94 21.00 21.55 2.97 12.0 26.0 14.0 -0.50 0.55 0.43
acsi_curr 25 46 17.26 4.61 17.00 16.97 5.93 10.5 26.0 15.5 0.41 -1.07 0.68
acsi_retain 26 46 80.67 18.24 82.50 81.26 22.98 42.0 110.0 68.0 -0.20 -1.08 2.69
acsl_prev 27 46 21.54 5.10 21.00 21.58 5.93 11.0 32.0 21.0 -0.01 -0.85 0.75
acsl_curr 28 46 17.18 5.39 16.00 16.92 5.56 8.5 28.0 19.5 0.43 -1.00 0.79
acsl_retain 29 46 80.48 18.46 79.00 80.68 17.79 44.0 122.0 78.0 -0.02 -0.52 2.72
acsf_prev 30 46 12.98 4.34 13.00 13.08 4.45 4.0 20.0 16.0 -0.22 -0.95 0.64
acsf_curr 31 46 8.33 3.93 8.25 8.14 3.34 1.0 18.0 17.0 0.32 -0.11 0.58
acsf_retain 32 46 67.00 33.19 63.00 63.68 22.98 9.0 200.0 191.0 1.54 3.98 4.89
acss_prev 33 46 13.33 1.51 13.50 13.42 1.48 9.0 17.0 8.0 -0.44 0.44 0.22
acss_curr 34 46 9.82 2.71 10.00 9.88 2.59 4.5 15.0 10.5 -0.26 -0.62 0.40
acss_retain 35 46 73.67 18.81 73.00 74.24 19.27 35.0 109.0 74.0 -0.20 -0.85 2.77
activity_card_sort_complete 36 46 2.00 0.00 2.00 2.00 0.00 2.0 2.0 0.0 NaN NaN 0.00
spstotal 37 46 79.35 11.29 82.00 79.87 14.83 55.0 96.0 41.0 -0.33 -1.19 1.67
bfi_extraversion 38 46 6.93 2.36 7.00 7.05 2.97 2.0 10.0 8.0 -0.23 -1.02 0.35
bfi_agreeable 39 46 7.13 2.02 7.00 7.21 2.97 3.0 10.0 7.0 -0.40 -0.92 0.30
bfi_consciousness 40 46 7.87 1.77 8.00 8.03 1.48 3.0 10.0 7.0 -0.70 -0.33 0.26
bfi_neuroticism 41 46 6.28 2.46 6.00 6.32 2.97 2.0 10.0 8.0 -0.02 -1.15 0.36
bfi_openness 42 46 7.85 2.02 8.00 8.03 2.97 2.0 10.0 8.0 -0.54 -0.41 0.30
frsbe_exec 43 46 41.83 10.09 43.00 41.53 10.38 24.0 63.0 39.0 0.11 -0.82 1.49
frsbe_apathy 44 46 33.28 8.61 32.00 32.82 10.38 18.0 53.0 35.0 0.43 -0.71 1.27
frsbe_disinhib 45 46 32.72 5.95 32.00 32.58 6.67 21.0 46.0 25.0 0.17 -0.77 0.88
frsbe_total 46 46 107.83 20.18 109.00 107.34 22.98 72.0 150.0 78.0 0.21 -0.89 2.97
frsbe_complete 47 46 2.00 0.00 2.00 2.00 0.00 2.0 2.0 0.0 NaN NaN 0.00
tbiqol_part_sra_tscore 48 46 46.33 6.70 46.00 45.81 5.78 32.1 64.1 32.0 0.83 1.10 0.99
tbiqol_anger_tscore 49 46 51.13 9.95 51.60 51.04 11.49 33.1 69.9 36.8 0.07 -1.04 1.47
tbiqol_anxiety_tscore 50 46 55.64 9.23 56.05 55.89 9.56 36.1 73.0 36.9 -0.23 -0.73 1.36
tbiqol_comm_tscore 51 46 46.25 9.56 45.55 46.14 8.97 29.2 65.5 36.3 0.15 -0.87 1.41
tbiqol_depression_tscore 52 46 53.95 9.78 53.85 54.10 10.16 33.6 74.0 40.4 -0.08 -0.72 1.44
tbiqol_dyscontrol_tscore 53 46 50.79 8.19 52.30 51.19 7.56 33.2 66.8 33.6 -0.42 -0.44 1.21
tbiqol_execfunc_tscore 54 46 35.48 6.06 34.30 35.22 5.34 24.3 50.8 26.5 0.38 -0.51 0.89
tbiqol_fatigue_tscore 55 46 54.99 8.59 54.65 55.06 8.23 37.9 72.5 34.6 0.01 -0.70 1.27
tbiqol_genconcern_tscore 56 46 36.16 8.80 35.85 35.99 8.90 19.7 53.8 34.1 0.18 -0.72 1.30
tbiqol_grief_tscore 57 46 52.53 9.40 53.65 53.07 7.26 30.7 70.3 39.6 -0.60 -0.12 1.39
tbiqol_headache_tscore 58 46 49.44 9.22 49.50 48.98 13.20 38.5 67.1 28.6 0.14 -1.33 1.36
tbiqol_mobility_tscore 59 46 45.69 8.75 44.40 45.26 8.45 31.5 63.6 32.1 0.46 -0.72 1.29
tbiqol_pain_tscore 60 46 55.26 10.73 57.05 55.24 11.49 38.4 74.8 36.4 -0.24 -1.09 1.58
tbiqol_posaffect_tscore 61 46 50.09 7.43 49.65 49.95 7.71 35.4 68.9 33.5 0.21 -0.51 1.10
tbiqol_resilience_tscore 62 46 48.45 8.09 49.10 48.12 7.78 33.4 73.6 40.2 0.44 0.43 1.19
tbiqol_selfesteem_tscore 63 46 48.06 10.73 48.45 48.10 10.75 28.4 66.0 37.6 -0.01 -0.97 1.58
tbiqol_satissra_tscore 64 46 45.23 6.25 45.10 44.79 4.60 34.7 63.2 28.5 0.88 1.18 0.92
tbiqol_stigma_tscore 65 45 50.98 7.39 52.00 51.61 6.08 33.5 62.3 28.8 -0.73 -0.12 1.10
tbiqol_ue_tscore 66 46 44.15 8.56 42.50 44.05 8.30 27.9 58.1 30.2 0.39 -0.91 1.26
marital_status 67 46 1.72 0.72 2.00 1.66 1.48 1.0 3.0 2.0 0.46 -1.03 0.11
phys_health 68 46 110.25 17.18 110.70 110.24 20.90 82.6 143.3 60.7 -0.06 -1.11 2.53
phys_health_index 69 46 94.57 13.85 95.00 94.74 16.31 64.0 117.0 53.0 -0.12 -0.98 2.04
emo_health 70 46 160.72 25.68 158.45 161.31 31.65 115.1 201.3 86.2 -0.10 -1.28 3.79
emo_health_index 71 46 99.33 14.10 101.00 99.11 17.79 77.0 123.0 46.0 0.05 -1.35 2.08
cog_health 72 46 71.64 14.36 70.35 71.32 14.53 44.0 104.6 60.6 0.28 -0.61 2.12
cog_health_index 73 46 94.11 14.73 93.00 94.13 16.31 62.0 123.0 61.0 0.02 -0.75 2.17
soc_health 74 46 91.56 12.05 90.15 90.89 10.38 69.6 127.3 57.7 0.67 0.59 1.78
soc_health_index 75 46 92.65 12.23 92.50 92.89 10.38 64.0 122.0 58.0 -0.16 0.27 1.80
glob_health 76 46 380.65 45.57 380.00 379.84 55.60 303.0 468.0 165.0 0.08 -1.13 6.72
glob_health_index 77 46 94.37 13.44 95.00 94.11 15.57 71.0 120.0 49.0 0.04 -1.06 1.98

ACS3

Outcome variable: ACS3 for all scores mean(sd)

Characteristic N = 461
ACS Global Before 69 (10)
ACS Global Current 53 (14)
Global Retained (%) 76 (16)
ACS IADL Before 21.46 (2.94)
ACS IADL Current 17.3 (4.6)
IADL Retained (%) 81 (18)
ACS Leisure Before 21.5 (5.1)
ACS Leisure Current 17.2 (5.4)
Leisure Retained (%) 80 (18)
ACS Fitness Before 13.0 (4.3)
ACS Fitness Current 8.3 (3.9)
Fitness Retained (%) 67 (33)
ACS Social Before 13.33 (1.51)
ACS Social Current 9.82 (2.71)
Social Retained (%) 74 (19)
1 Mean (SD)

Correlations

All variables

Correlation of all variables of interest with TBI QOL subscores. While too small to read in HTML print out, nice reference during analysis

Correlation matrix of PPF only (using composite TBIQOL Scores)

Included Variables

Matrix with heat map for all included variables in dissertation. Figure 4 in dissertation

Below is the breakdown of all TBIQOL sub scores with the ACS3. While not included in this study, helpful for discussion and future publications.

FrSBe and TBI-QOL

TBI QOL subscales with the FrSBe

RQ1: Regression Analysis

Research Question 1 1. What is the relationship between protective factors and self-regulation with resiliency-related outcomes such as re-engagement in meaningful activities? a. To what extent do protective factors and self-regulation predict resiliency-related outcomes in the TBI population? Hypothesis: Higher self-regulation will be associated with better resiliency-related outcomes b. To what extent does self-regulation mediate or moderate the influence of protective factors on resiliency-related outcomes after TBI? Hypothesis: Self-regulation will impact the relationship between protective factors and resiliency-related outcomes

RQ1a

First, we’ll look at the hierarchical linear model as outlined in Chapter 3. Then, to dive deeper, a “post hoc” analysis of each subscale of the ACS and use AIC to determine model of best fit.

ACS Global

In this section, we’ll do the original hierarchical model with protective and environmental protective factors in the first step and then total self-regulation score added for the second. *note that the cognitive composite score is not included as it includes exec functioning, which in this paper is seen as a self-regulatory process. therefore, gen concerns (memory and concentration) is used as cognitive protective factor

step1 <- lm(acsg_retain~age_current, data=df)
summary(step1)
## 
## Call:
## lm(formula = acsg_retain ~ age_current, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.645 -11.542   0.093   7.501  36.764 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  92.7375     8.0352   11.54  6.7e-15 ***
## age_current  -0.3561     0.1634   -2.18   0.0347 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.82 on 44 degrees of freedom
## Multiple R-squared:  0.09745,    Adjusted R-squared:  0.07693 
## F-statistic: 4.751 on 1 and 44 DF,  p-value: 0.03468
step2<- lm(acsg_retain~age_current+phys_health_index+tbiqol_genconcern_tscore+emo_health_index+ spstotal, data=df)
summary(step2)
## 
## Call:
## lm(formula = acsg_retain ~ age_current + phys_health_index + 
##     tbiqol_genconcern_tscore + emo_health_index + spstotal, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -25.997 -10.216  -1.356  10.269  30.600 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)  
## (Intercept)              43.23068   25.86980   1.671   0.1025  
## age_current              -0.13302    0.16854  -0.789   0.4346  
## phys_health_index         0.07379    0.23057   0.320   0.7506  
## tbiqol_genconcern_tscore  0.80590    0.37270   2.162   0.0366 *
## emo_health_index         -0.09498    0.22396  -0.424   0.6738  
## spstotal                  0.15527    0.24641   0.630   0.5322  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 14.7 on 40 degrees of freedom
## Multiple R-squared:  0.2916, Adjusted R-squared:  0.203 
## F-statistic: 3.293 on 5 and 40 DF,  p-value: 0.0139
#Nested Model Comparison
anova(step1, step2)
## Analysis of Variance Table
## 
## Model 1: acsg_retain ~ age_current
## Model 2: acsg_retain ~ age_current + phys_health_index + tbiqol_genconcern_tscore + 
##     emo_health_index + spstotal
##   Res.Df     RSS Df Sum of Sq      F  Pr(>F)  
## 1     44 11012.0                              
## 2     40  8643.4  4    2368.7 2.7404 0.04176 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#change in R-squared
summary(step2)$r.squared - summary(step1)$r.squared
## [1] 0.1941371
step3<- lm(acsg_retain~age_current+phys_health_index+tbiqol_genconcern_tscore+emo_health_index+ spstotal+frsbe_total, data=df)
summary(step3)
## 
## Call:
## lm(formula = acsg_retain ~ age_current + phys_health_index + 
##     tbiqol_genconcern_tscore + emo_health_index + spstotal + 
##     frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -25.765  -9.412  -1.115   9.607  30.674 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)  
## (Intercept)              54.89809   39.18755   1.401   0.1691  
## age_current              -0.13795    0.17078  -0.808   0.4241  
## phys_health_index         0.07706    0.23318   0.330   0.7428  
## tbiqol_genconcern_tscore  0.75110    0.40085   1.874   0.0685 .
## emo_health_index         -0.11110    0.22992  -0.483   0.6316  
## spstotal                  0.13131    0.25616   0.513   0.6111  
## frsbe_total              -0.05806    0.14525  -0.400   0.6916  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 14.86 on 39 degrees of freedom
## Multiple R-squared:  0.2945, Adjusted R-squared:  0.1859 
## F-statistic: 2.713 on 6 and 39 DF,  p-value: 0.02676
#Nested Model Comparison
anova(step2, step3)
## Analysis of Variance Table
## 
## Model 1: acsg_retain ~ age_current + phys_health_index + tbiqol_genconcern_tscore + 
##     emo_health_index + spstotal
## Model 2: acsg_retain ~ age_current + phys_health_index + tbiqol_genconcern_tscore + 
##     emo_health_index + spstotal + frsbe_total
##   Res.Df    RSS Df Sum of Sq      F Pr(>F)
## 1     40 8643.4                           
## 2     39 8608.1  1    35.263 0.1598 0.6916
#change in R-squared
summary(step3)$r.squared - summary(step2)$r.squared
## [1] 0.002890204

Assumptions tested

Test for multicollinearity

##              age_current        phys_health_index tbiqol_genconcern_tscore 
##                 1.239106                 2.126709                 2.538554 
##         emo_health_index                 spstotal              frsbe_total 
##                 2.142136                 1.706537                 1.751156

Post Hoc AIC Models

As we have a smaller n and need to be parsimonious with the variables we use in the regression model, we’ll look at several models based on correlations (higher correlations added to the models) and then calculate the AIC. The lower AIC, the better the fit and that model will be used

Note that all assumptions were tested for each model and were met

ACS3 Social

## 
## Model selection based on AICc:
## 
##         K   AICc Delta_AICc AICcWt Cum.Wt      LL
## model7  7 390.45       0.00   0.39   0.39 -186.75
## model11 8 391.69       1.24   0.21   0.60 -185.90
## model6  8 393.39       2.94   0.09   0.68 -186.75
## model9  7 393.57       3.12   0.08   0.77 -188.31
## model10 7 393.57       3.12   0.08   0.85 -188.31
## model8  7 395.18       4.73   0.04   0.88 -189.12
## model3  7 395.23       4.78   0.04   0.92 -189.14
## model4  7 395.37       4.92   0.03   0.95 -189.21
## model1  6 395.52       5.07   0.03   0.98 -190.68
## model5  8 397.94       7.50   0.01   0.99 -189.03
## model2  7 398.02       7.58   0.01   1.00 -190.54
## 
## Call:
## lm(formula = acss_retain ~ tbiqol_fatigue_tscore + tbiqol_genconcern_tscore + 
##     tbiqol_depression_tscore + tbiqol_anxiety_tscore + frsbe_apathy, 
##     data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.866  -9.287   0.797  12.256  23.790 
## 
## Coefficients:
##                           Estimate Std. Error t value Pr(>|t|)   
## (Intercept)              100.44884   31.00124   3.240  0.00241 **
## tbiqol_fatigue_tscore      0.05358    0.37022   0.145  0.88566   
## tbiqol_genconcern_tscore   0.63278    0.34196   1.850  0.07164 . 
## tbiqol_depression_tscore   0.37030    0.35414   1.046  0.30201   
## tbiqol_anxiety_tscore     -0.87342    0.35016  -2.494  0.01685 * 
## frsbe_apathy              -0.72065    0.32810  -2.196  0.03391 * 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.04 on 40 degrees of freedom
## Multiple R-squared:  0.4316, Adjusted R-squared:  0.3606 
## F-statistic: 6.076 on 5 and 40 DF,  p-value: 0.0002835

ACS3 IADL

## 
## Model selection based on AICc:
## 
##        K   AICc Delta_AICc AICcWt Cum.Wt      LL
## model6 7 397.09       0.00   0.35   0.35 -190.07
## model4 7 397.61       0.51   0.27   0.63 -190.33
## model3 8 399.11       2.01   0.13   0.76 -189.61
## model5 8 399.87       2.77   0.09   0.85 -189.99
## model2 8 400.09       3.00   0.08   0.93 -190.10
## model1 8 400.25       3.15   0.07   1.00 -190.18

ACS3 Leisure

## 
## Model selection based on AICc:
## 
##        K   AICc Delta_AICc AICcWt Cum.Wt      LL
## model7 6 397.56       0.00   0.28   0.28 -191.70
## model6 6 397.58       0.02   0.28   0.57 -191.71
## model3 6 397.88       0.32   0.24   0.81 -191.86
## model1 7 399.92       2.35   0.09   0.90 -191.48
## model4 7 400.36       2.80   0.07   0.97 -191.71
## model2 8 402.73       5.17   0.02   0.99 -191.42
## model5 7 404.08       6.52   0.01   1.00 -193.57
## 
## Call:
## lm(formula = acsl_retain ~ tbiqol_mobility_tscore + tbiqol_genconcern_tscore + 
##     tbiqol_comm_tscore + frsbe_apathy, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -34.458 -10.970  -0.449  10.394  35.565 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)  
## (Intercept)               44.1102    23.5848   1.870   0.0686 .
## tbiqol_mobility_tscore     0.6630     0.3502   1.893   0.0654 .
## tbiqol_genconcern_tscore   0.8003     0.3665   2.184   0.0347 *
## tbiqol_comm_tscore        -0.3487     0.3847  -0.906   0.3700  
## frsbe_apathy              -0.2025     0.3330  -0.608   0.5466  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 16.55 on 41 degrees of freedom
## Multiple R-squared:  0.2675, Adjusted R-squared:  0.196 
## F-statistic: 3.743 on 4 and 41 DF,  p-value: 0.01098

ACS3 Fitness

## 
## Model selection based on AICc:
## 
##        K   AICc Delta_AICc AICcWt Cum.Wt      LL
## model2 6 450.93       0.00   0.81   0.81 -218.36
## model1 8 455.12       4.19   0.10   0.91 -217.56
## model5 8 455.54       4.62   0.08   0.99 -217.77
## model4 6 459.70       8.77   0.01   1.00 -222.77
## model3 7 461.23      10.30   0.00   1.00 -222.14
## 
## Call:
## lm(formula = acsf_retain ~ phys_health_index + tbiqol_genconcern_tscore + 
##     tbiqol_anxiety_tscore + tbiqol_stigma_tscore, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -61.962 -17.262  -7.947  11.301 124.325 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)
## (Intercept)               70.6891    70.1111   1.008    0.319
## phys_health_index          0.3343     0.4807   0.695    0.491
## tbiqol_genconcern_tscore   0.3498     0.7653   0.457    0.650
## tbiqol_anxiety_tscore     -0.1997     0.7241  -0.276    0.784
## tbiqol_stigma_tscore      -0.7217     0.8918  -0.809    0.423
## 
## Residual standard error: 32.86 on 40 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.1283, Adjusted R-squared:  0.04115 
## F-statistic: 1.472 on 4 and 40 DF,  p-value: 0.2288

RQ1b

Moderation

For moderation, looking at personal protective factors and environmental protective factors from original model

## 
## Call:
## lm(formula = acsg_retain ~ phys_health_index * frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.431  -8.709  -1.636  10.093  33.238 
## 
## Coefficients:
##                                 Estimate Std. Error t value Pr(>|t|)
## (Intercept)                    18.928754 103.340073   0.183    0.856
## phys_health_index               0.815796   1.032293   0.790    0.434
## frsbe_total                     0.241520   0.923153   0.262    0.795
## phys_health_index:frsbe_total  -0.004570   0.009366  -0.488    0.628
## 
## Residual standard error: 15.34 on 42 degrees of freedom
## Multiple R-squared:  0.1898, Adjusted R-squared:  0.1319 
## F-statistic:  3.28 on 3 and 42 DF,  p-value: 0.03006
## 
## Call:
## lm(formula = acsg_retain ~ tbiqol_genconcern_tscore * frsbe_total, 
##     data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -28.3975  -9.9418   0.9413   8.4603  31.4643 
## 
## Coefficients:
##                                       Estimate Std. Error t value Pr(>|t|)
## (Intercept)                          43.336712  52.829896   0.820    0.417
## tbiqol_genconcern_tscore              1.080987   1.386257   0.780    0.440
## frsbe_total                           0.003913   0.476477   0.008    0.993
## tbiqol_genconcern_tscore:frsbe_total -0.001809   0.013296  -0.136    0.892
## 
## Residual standard error: 14.51 on 42 degrees of freedom
## Multiple R-squared:  0.2752, Adjusted R-squared:  0.2235 
## F-statistic: 5.317 on 3 and 42 DF,  p-value: 0.003381
## 
## Call:
## lm(formula = acsg_retain ~ emo_health_index * frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.984  -7.556   0.125   9.355  30.142 
## 
## Coefficients:
##                                Estimate Std. Error t value Pr(>|t|)
## (Intercept)                  -76.646620 110.391140  -0.694    0.491
## emo_health_index               1.681344   1.027245   1.637    0.109
## frsbe_total                    1.182751   0.955996   1.237    0.223
## emo_health_index:frsbe_total  -0.013429   0.009071  -1.480    0.146
## 
## Residual standard error: 15.39 on 42 degrees of freedom
## Multiple R-squared:  0.1842, Adjusted R-squared:  0.1259 
## F-statistic:  3.16 on 3 and 42 DF,  p-value: 0.03435
## 
## Call:
## lm(formula = acsg_retain ~ spstotal * frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -32.255  -8.786   0.385   7.966  29.966 
## 
## Coefficients:
##                       Estimate Std. Error t value Pr(>|t|)
## (Intercept)           36.98213  115.10290   0.321    0.750
## spstotal               0.78362    1.42193   0.551    0.584
## frsbe_total            0.17807    1.04175   0.171    0.865
## spstotal:frsbe_total  -0.00502    0.01318  -0.381    0.705
## 
## Residual standard error: 15.74 on 42 degrees of freedom
## Multiple R-squared:  0.1476, Adjusted R-squared:  0.08675 
## F-statistic: 2.425 on 3 and 42 DF,  p-value: 0.07901

There was no moderating effect of apathy on any of the predictors.

Mediation

Here we look at mediation effect of the total FrSBe scores on the personal and environmental factors used in the post hoc model (mobility, general cog functioning, anxiety, depression, and social support)

How to read: ACME = indirect effect ADE = direct effect ACME + ADE = total effect

Physical Health

# Initial Model
model1 <- lm(acsg_retain ~ phys_health_index, df) # Y ~ X, DV predicted by IV - no mediation considered - total effect
summary(model1)
## 
## Call:
## lm(formula = acsg_retain ~ phys_health_index, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -32.165  -8.827  -2.595  11.831  33.431 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)  
## (Intercept)        35.3344    15.9664   2.213   0.0321 *
## phys_health_index   0.4298     0.1671   2.572   0.0136 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.53 on 44 degrees of freedom
## Multiple R-squared:  0.1307, Adjusted R-squared:  0.111 
## F-statistic: 6.616 on 1 and 44 DF,  p-value: 0.01356
# Mediation paths
medmodel1 <- lm(frsbe_total ~ phys_health_index, df) # M ~ X, mediator predicted by X
outputmodel1 <- lm(acsg_retain ~ phys_health_index + frsbe_total, df) # Y ~ X + M, DV predicted by mediator, adjusting for IV

summary(medmodel1)
## 
## Call:
## lm(formula = frsbe_total ~ phys_health_index, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -32.904 -14.259  -1.361  12.529  48.322 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       158.6667    19.5030   8.135 2.58e-10 ***
## phys_health_index  -0.5376     0.2041  -2.634   0.0116 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 18.96 on 44 degrees of freedom
## Multiple R-squared:  0.1362, Adjusted R-squared:  0.1166 
## F-statistic: 6.938 on 1 and 44 DF,  p-value: 0.0116
summary(outputmodel1)
## 
## Call:
## lm(formula = acsg_retain ~ phys_health_index + frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -31.877  -8.530  -0.878   9.844  34.180 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)   
## (Intercept)        67.8622    24.7442   2.743  0.00885 **
## phys_health_index   0.3196     0.1761   1.815  0.07649 . 
## frsbe_total        -0.2050     0.1209  -1.696  0.09709 . 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.2 on 43 degrees of freedom
## Multiple R-squared:  0.1852, Adjusted R-squared:  0.1473 
## F-statistic: 4.888 on 2 and 43 DF,  p-value: 0.01223
# Mediation test
mediation <- mediate(medmodel1, # Mediator model
                    outputmodel1, # Outcome model
                    boot = T, # Ask for bootstrapped confidence intervals
                    treat="phys_health_index", # Name of the x variable
                    mediator="frsbe_total" # Name of the m variable
                    )
# if you don't want bootstrap, just delete 'sims' line and set boot = F

summary(mediation)
## 
## Causal Mediation Analysis 
## 
## Nonparametric Bootstrap Confidence Intervals with the Percentile Method
## 
##                Estimate 95% CI Lower 95% CI Upper p-value   
## ACME            0.11022      0.00332         0.28   0.036 * 
## ADE             0.31958     -0.02036         0.67   0.058 . 
## Total Effect    0.42980      0.08837         0.77   0.006 **
## Prop. Mediated  0.25644      0.00937         1.06   0.042 * 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 46 
## 
## 
## Simulations: 1000
plot(mediation)

There is a significant indirect effect and an insignificant direct effect, indicating total mediation

## [1] "-0.54"
## [1] "-0.21"

#### Cognitive Health

## 
## Call:
## lm(formula = acsg_retain ~ tbiqol_genconcern_tscore, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -28.3103 -10.7011   0.1933   9.3422  31.1265 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)    
## (Intercept)               40.7291     8.9525   4.549  4.2e-05 ***
## tbiqol_genconcern_tscore   0.9747     0.2407   4.050 0.000205 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 14.21 on 44 degrees of freedom
## Multiple R-squared:  0.2715, Adjusted R-squared:  0.255 
## F-statistic:  16.4 on 1 and 44 DF,  p-value: 0.0002048
## 
## Call:
## lm(formula = frsbe_total ~ tbiqol_genconcern_tscore, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -33.175  -9.996   0.624  10.540  40.890 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)    
## (Intercept)              156.0357    10.4557  14.923  < 2e-16 ***
## tbiqol_genconcern_tscore  -1.3331     0.2811  -4.743 2.25e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 16.6 on 44 degrees of freedom
## Multiple R-squared:  0.3383, Adjusted R-squared:  0.3232 
## F-statistic: 22.49 on 1 and 44 DF,  p-value: 2.245e-05
## 
## Call:
## lm(formula = acsg_retain ~ tbiqol_genconcern_tscore + frsbe_total, 
##     data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -28.5347 -10.1387   0.7124   8.4955  31.4666 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)   
## (Intercept)               49.8410    22.2441   2.241  0.03027 * 
## tbiqol_genconcern_tscore   0.8969     0.2986   3.004  0.00443 **
## frsbe_total               -0.0584     0.1303  -0.448  0.65621   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 14.34 on 43 degrees of freedom
## Multiple R-squared:  0.2749, Adjusted R-squared:  0.2412 
## F-statistic: 8.152 on 2 and 43 DF,  p-value: 0.0009958
## 
## Causal Mediation Analysis 
## 
## Nonparametric Bootstrap Confidence Intervals with the Percentile Method
## 
##                Estimate 95% CI Lower 95% CI Upper p-value    
## ACME             0.0778      -0.1526         0.34   0.468    
## ADE              0.8969       0.3873         1.37   0.002 ** 
## Total Effect     0.9747       0.5170         1.47  <2e-16 ***
## Prop. Mediated   0.0799      -0.1947         0.38   0.468    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 46 
## 
## 
## Simulations: 1000

There is no significant indirect effect. No mediation

Emotional Health

## 
## Call:
## lm(formula = acsg_retain ~ emo_health_index, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -35.342  -9.104  -0.366   9.243  31.047 
## 
## Coefficients:
##                  Estimate Std. Error t value Pr(>|t|)  
## (Intercept)       41.4375    16.8596   2.458   0.0180 *
## emo_health_index   0.3478     0.1681   2.069   0.0445 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.9 on 44 degrees of freedom
## Multiple R-squared:  0.08865,    Adjusted R-squared:  0.06794 
## F-statistic:  4.28 on 1 and 44 DF,  p-value: 0.04447
## 
## Call:
## lm(formula = frsbe_total ~ emo_health_index, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -27.595 -12.448  -1.991   8.070  48.441 
## 
## Coefficients:
##                  Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      180.9026    18.5626   9.746 1.47e-12 ***
## emo_health_index  -0.7357     0.1851  -3.975 0.000258 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.5 on 44 degrees of freedom
## Multiple R-squared:  0.2643, Adjusted R-squared:  0.2475 
## F-statistic:  15.8 on 1 and 44 DF,  p-value: 0.0002579
## 
## Call:
## lm(formula = acsg_retain ~ emo_health_index + frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -33.527  -8.166   0.086   7.921  31.359 
## 
## Coefficients:
##                  Estimate Std. Error t value Pr(>|t|)   
## (Intercept)       81.0401    29.4160   2.755  0.00857 **
## emo_health_index   0.1867     0.1924   0.970  0.33728   
## frsbe_total       -0.2189     0.1344  -1.629  0.11071   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.61 on 43 degrees of freedom
## Multiple R-squared:  0.1416, Adjusted R-squared:  0.1017 
## F-statistic: 3.546 on 2 and 43 DF,  p-value: 0.03753
## 
## Causal Mediation Analysis 
## 
## Nonparametric Bootstrap Confidence Intervals with the Percentile Method
## 
##                Estimate 95% CI Lower 95% CI Upper p-value  
## ACME            0.16106     -0.02350         0.34   0.084 .
## ADE             0.18669     -0.22926         0.63   0.334  
## Total Effect    0.34775      0.00944         0.71   0.046 *
## Prop. Mediated  0.46315     -0.37071         3.17   0.130  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 46 
## 
## 
## Simulations: 1000

There is no significant indirect effect indicating no mediation

## [1] "-0.74"
## [1] "-0.22"

#### SPS

## 
## Call:
## lm(formula = acsg_retain ~ spstotal, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -34.044  -8.542   0.400   8.817  31.509 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  40.7396    16.7745   2.429   0.0193 *
## spstotal      0.4441     0.2093   2.121   0.0396 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.86 on 44 degrees of freedom
## Multiple R-squared:  0.09279,    Adjusted R-squared:  0.07218 
## F-statistic: 4.501 on 1 and 44 DF,  p-value: 0.03955
## 
## Call:
## lm(formula = frsbe_total ~ spstotal, data = df)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -38.16 -12.89  -1.67  10.30  36.27 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 179.3888    18.6288   9.630 2.11e-12 ***
## spstotal     -0.9019     0.2325  -3.879 0.000346 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.61 on 44 degrees of freedom
## Multiple R-squared:  0.2549, Adjusted R-squared:  0.2379 
## F-statistic: 15.05 on 1 and 44 DF,  p-value: 0.0003463
## 
## Call:
## lm(formula = acsg_retain ~ spstotal + frsbe_total, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -32.888  -8.280   0.220   7.625  29.910 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)  79.3728    29.0439   2.733  0.00908 **
## spstotal      0.2499     0.2382   1.049  0.30003   
## frsbe_total  -0.2154     0.1333  -1.615  0.11358   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.58 on 43 degrees of freedom
## Multiple R-squared:  0.1447, Adjusted R-squared:  0.1049 
## F-statistic: 3.637 on 2 and 43 DF,  p-value: 0.03473
## 
## Causal Mediation Analysis 
## 
## Nonparametric Bootstrap Confidence Intervals with the Percentile Method
## 
##                 Estimate 95% CI Lower 95% CI Upper p-value  
## ACME            0.194230     0.000756         0.44   0.050 *
## ADE             0.249874    -0.223153         0.73   0.294  
## Total Effect    0.444104     0.035102         0.89   0.032 *
## Prop. Mediated  0.437353    -0.056627         2.71   0.082 .
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 46 
## 
## 
## Simulations: 1000

There is a significant indirect effect and insignificant direct effect indicating full mediation

## [1] "-0.9"
## [1] "-0.22"

RQ2: Time Since Injury

  1. How does time since injury influence resiliency?
  1. Is time after injury associated with an individual’s ability to re-engage in meaningful activities?
  2. To what extent do the relationships between protective factors and self-regulatory processes with resiliency-related outcomes change with time since sustaining TBI? Hypothesis: Time since injury will be associated with resiliency, leading to different resiliency-related outcomes

To answer these questions, first look at descriptive statistics, then regression model with time since injury included, lastly, investigate what, if any, role time since injury has on protective factors and self-regulation

Descriptives

##             Min. 1st Qu. Median Mean 3rd Qu. Max.
## time_injury    1     3.5      7 9.09      11   30

Correlation ACS

Regression Analysis

Global ACS

First, looking just at the relationship between time since injury and re-engagement while controlling for age.

## 
## Call:
## lm(formula = acsg_retain ~ age_current + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -27.815 -10.656  -1.068   6.706  40.845 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  89.6687     8.0476  11.142 2.93e-14 ***
## age_current  -0.4019     0.1618  -2.484   0.0170 *  
## time_injury   0.5751     0.3290   1.748   0.0876 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.46 on 43 degrees of freedom
## Multiple R-squared:  0.1573, Adjusted R-squared:  0.1181 
## F-statistic: 4.014 on 2 and 43 DF,  p-value: 0.02521

Controlling for age, there is no significant relationship between time since injury and re-engagement

Social ACS3

## 
## Call:
## lm(formula = acss_retain ~ age_current + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -39.707 -12.341   3.187  13.229  26.463 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  87.0980     9.0877   9.584 3.08e-12 ***
## age_current  -0.4428     0.1827  -2.423   0.0197 *  
## time_injury   0.8160     0.3715   2.197   0.0335 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.46 on 43 degrees of freedom
## Multiple R-squared:  0.1765, Adjusted R-squared:  0.1382 
## F-statistic: 4.607 on 2 and 43 DF,  p-value: 0.01539

IADL ACS3

## 
## Call:
## lm(formula = acsi_retain ~ age_current + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -34.001 -14.491  -1.073  15.727  31.400 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  90.2276     9.2527   9.751 1.84e-12 ***
## age_current  -0.3122     0.1861  -1.678    0.101    
## time_injury   0.5659     0.3782   1.496    0.142    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.78 on 43 degrees of freedom
## Multiple R-squared:  0.09199,    Adjusted R-squared:  0.04975 
## F-statistic: 2.178 on 2 and 43 DF,  p-value: 0.1256

Leisure ACS3

## 
## Call:
## lm(formula = acsl_retain ~ age_current + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -28.517 -12.231   0.136  12.241  53.013 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  96.6837     9.1006  10.624 1.33e-13 ***
## age_current  -0.4482     0.1830  -2.449   0.0185 *  
## time_injury   0.5379     0.3720   1.446   0.1555    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.49 on 43 degrees of freedom
## Multiple R-squared:  0.1422, Adjusted R-squared:  0.1023 
## F-statistic: 3.564 on 2 and 43 DF,  p-value: 0.03697

Fitness ACS3

## 
## Call:
## lm(formula = acsf_retain ~ age_current + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -48.889 -18.546  -5.539   9.106 129.208 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  80.4566    16.9992   4.733 2.41e-05 ***
## age_current  -0.4788     0.3418  -1.401    0.168    
## time_injury   0.9989     0.6949   1.438    0.158    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 32.66 on 43 degrees of freedom
## Multiple R-squared:  0.07462,    Adjusted R-squared:  0.03158 
## F-statistic: 1.734 on 2 and 43 DF,  p-value: 0.1888

Hierarchical Regression: Step 4

## 
## Call:
## lm(formula = acsg_retain ~ age_current + phys_health_index + 
##     tbiqol_genconcern_tscore + emo_health_index + spstotal + 
##     frsbe_total + time_injury, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -25.772  -7.984  -2.549   7.321  30.924 
## 
## Coefficients:
##                          Estimate Std. Error t value Pr(>|t|)  
## (Intercept)               49.6084    36.9543   1.342   0.1874  
## age_current               -0.1810     0.1617  -1.119   0.2700  
## phys_health_index          0.1234     0.2203   0.560   0.5786  
## tbiqol_genconcern_tscore   0.6892     0.3782   1.822   0.0763 .
## emo_health_index          -0.1499     0.2170  -0.691   0.4941  
## spstotal                   0.2162     0.2436   0.887   0.3804  
## frsbe_total               -0.1001     0.1378  -0.726   0.4722  
## time_injury                0.7499     0.3060   2.451   0.0190 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13.99 on 38 degrees of freedom
## Multiple R-squared:  0.3908, Adjusted R-squared:  0.2785 
## F-statistic: 3.482 on 7 and 38 DF,  p-value: 0.005578
#Nested Model Comparison
anova(step3, step4)
## Analysis of Variance Table
## 
## Model 1: acsg_retain ~ age_current + phys_health_index + tbiqol_genconcern_tscore + 
##     emo_health_index + spstotal + frsbe_total
## Model 2: acsg_retain ~ age_current + phys_health_index + tbiqol_genconcern_tscore + 
##     emo_health_index + spstotal + frsbe_total + time_injury
##   Res.Df    RSS Df Sum of Sq      F  Pr(>F)  
## 1     39 8608.1                              
## 2     38 7433.2  1    1174.9 6.0062 0.01897 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#change in R-squared
summary(step4)$r.squared - summary(step3)$r.squared
## [1] 0.09629406

POST HOC: Groups

Because there was no significant relationship between time since injury and outcomes BUT it was a significant predictor in the model, I seperated time since recovery into three groups to examine relationships further. Early = 6mo-3years Mid = 3 to 10 years Later = >10 years

1= everything <=3 and 3 is everything >=10

Characteristic Early, N = 111 Mid, N = 141 Later, N = 211
Age (years) 41 (13) 50 (17) 48 (13)
Employment status


    0 4 (36%) 11 (79%) 12 (57%)
    1 7 (64%) 3 (21%) 9 (43%)
Substance use score 1.82 (1.66) 3.21 (3.53) 2.81 (3.16)
Severity of Injury


    2 5 (45%) 7 (50%) 7 (33%)
    3 6 (55%) 7 (50%) 14 (67%)
Global ACS 71 (14) 73 (18) 80 (16)
Social ACS 70 (17) 69 (14) 79 (22)
IADL ACS 74 (19) 79 (20) 85 (16)
Leisure ACS 74 (17) 79 (21) 85 (17)
Fitness ACS 61 (21) 61 (31) 74 (40)
1 Mean (SD); n (%)
## 
## Early   Mid Later 
##    11    14    21

We see the counts of # participants in each group

Global

Global ACS3 scores (ie, global re-engagement scores)

## # A tibble: 3 × 4
##   time_injury_ex count  mean    sd
##   <fct>          <int> <dbl> <dbl>
## 1 Early             11  70.9  13.9
## 2 Mid               14  73.4  18.3
## 3 Later             21  80.4  16.0

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2    786   393.0    1.48  0.239
## Residuals      43  11415   265.5
## 
## Call:
## lm(formula = acsg_retain ~ age_current + time_injury_ex, data = df_time3)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -28.113 -11.124  -0.222   9.180  39.339 
## 
## Coefficients:
##                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          88.0693     8.1159  10.851 9.25e-14 ***
## age_current          -0.4204     0.1635  -2.572   0.0137 *  
## time_injury_exMid     6.3382     6.3566   0.997   0.3244    
## time_injury_exLater  12.6313     5.8341   2.165   0.0361 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.32 on 42 degrees of freedom
## Multiple R-squared:  0.1917, Adjusted R-squared:  0.134 
## F-statistic:  3.32 on 3 and 42 DF,  p-value: 0.02874

Controlling for age, we see a significant relationship between time since injury and global re engagement- specifically between early and later recovery

Social

social ACS3 scores (ie, social re-engagement scores)

## # A tibble: 3 × 4
##   time_injury_ex count  mean    sd
##   <fct>          <int> <dbl> <dbl>
## 1 Early             11  69.8  17.3
## 2 Mid               14  69.4  14.2
## 3 Later             21  78.5  21.7

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2    910   454.9   1.303  0.282
## Residuals      43  15010   349.1
## 
## Call:
## lm(formula = acss_retain ~ age_current + time_injury_ex, data = df_time3)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -37.734 -12.323   4.086  14.553  28.175 
## 
## Coefficients:
##                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          87.2722     9.4616   9.224 1.18e-11 ***
## age_current          -0.4276     0.1906  -2.244   0.0302 *  
## time_injury_exMid     3.5671     7.4106   0.481   0.6328    
## time_injury_exLater  11.9191     6.8015   1.752   0.0870 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.86 on 42 degrees of freedom
## Multiple R-squared:  0.1581, Adjusted R-squared:  0.09793 
## F-statistic: 2.628 on 3 and 42 DF,  p-value: 0.06263

IADL

IADL ACS3 scores (ie, IADL re-engagement scores)

## # A tibble: 3 × 4
##   time_injury_ex count  mean    sd
##   <fct>          <int> <dbl> <dbl>
## 1 Early             11  73.5  19.1
## 2 Mid               14  79.1  20.2
## 3 Later             21  85.4  15.8

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2   1067   533.3   1.649  0.204
## Residuals      43  13902   323.3
## 
## Call:
## lm(formula = acsi_retain ~ age_current + time_injury_ex, data = df_time3)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -28.0032 -14.6952  -0.5134  14.7014  29.3986 
## 
## Coefficients:
##                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          87.7460     9.2599   9.476 5.46e-12 ***
## age_current          -0.3479     0.1865  -1.865   0.0691 .  
## time_injury_exMid     8.8166     7.2526   1.216   0.2309    
## time_injury_exLater  14.4976     6.6564   2.178   0.0351 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.48 on 42 degrees of freedom
## Multiple R-squared:  0.1423, Adjusted R-squared:  0.08104 
## F-statistic: 2.323 on 3 and 42 DF,  p-value: 0.08878

We see a significant difference between early and late groups with IADl engagement when controlling for age

Leisure

Leisure ACS3 scores (ie, leisure re-engagement scores)

## # A tibble: 3 × 4
##   time_injury_ex count  mean    sd
##   <fct>          <int> <dbl> <dbl>
## 1 Early             11  74.3  17.0
## 2 Mid               14  78.8  21.0
## 3 Later             21  84.9  17.0

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2    866   433.2   1.288  0.286
## Residuals      43  14461   336.3
## 
## Call:
## lm(formula = acsl_retain ~ age_current + time_injury_ex, data = df_time3)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -30.295 -14.417   0.274  10.861  50.959 
## 
## Coefficients:
##                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          94.1180     9.0946  10.349 3.99e-13 ***
## age_current          -0.4862     0.1832  -2.654   0.0112 *  
## time_injury_exMid     9.0118     7.1232   1.265   0.2128    
## time_injury_exLater  14.2382     6.5377   2.178   0.0351 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 17.17 on 42 degrees of freedom
## Multiple R-squared:  0.192,  Adjusted R-squared:  0.1343 
## F-statistic: 3.328 on 3 and 42 DF,  p-value: 0.02851

We see a significant difference between early and late groups with Leisure engagement when controlling for age

Fitness

Fitness ACS3 scores (ie, fitness re-engagement scores)

## # A tibble: 3 × 4
##   time_injury_ex count  mean    sd
##   <fct>          <int> <dbl> <dbl>
## 1 Early             11  60.7  20.6
## 2 Mid               14  61.4  30.6
## 3 Later             21  74.0  39.5

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2   1922   960.8   0.867  0.427
## Residuals      43  47652  1108.2
## 
## Call:
## lm(formula = acsf_retain ~ age_current + time_injury_ex, data = df_time3)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -55.817 -19.104  -6.348   9.571 133.305 
## 
## Coefficients:
##                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          79.8850    17.4740   4.572 4.22e-05 ***
## age_current          -0.4693     0.3520  -1.334    0.190    
## time_injury_exMid     4.9728    13.6861   0.363    0.718    
## time_injury_exLater  16.8475    12.5612   1.341    0.187    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 32.99 on 42 degrees of freedom
## Multiple R-squared:  0.07781,    Adjusted R-squared:  0.01194 
## F-statistic: 1.181 on 3 and 42 DF,  p-value: 0.3284

Impact on Predictive Variables

This is exploratory and post hoc analysis- likely unable to report to avoid p-hacking, more for information gathering

###FrSBe

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2     19     9.5   0.022  0.978
## Residuals      43  18302   425.6

Physical Health

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2    219   109.3   0.559  0.576
## Residuals      43   8415   195.7

Emotional Health

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2    384   191.9   0.964  0.389
## Residuals      43   8560   199.1

General Cognition

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2     25   12.48   0.155  0.857
## Residuals      43   3462   80.51

Social Support

##                Df Sum Sq Mean Sq F value Pr(>F)
## time_injury_ex  2     25   12.48   0.155  0.857
## Residuals      43   3462   80.51

Cont Exploratory Analysis

Below is the comparison of t-values for step 3 and step 4 to see if the inclusion of time since injury significantly changed the contribution of each variable in the model. (not sure this is an appropriate way to report in paper, but wanted to see…)

##                   Variable     z_score   p_value
## 1              age_current  1.32457593 0.1853118
## 2        phys_health_index -0.71621797 0.4738568
## 3         emo_health_index  0.65568756 0.5120252
## 4 tbiqol_genconcern_tscore  0.09340553 0.9255814
## 5                 spstotal -1.06041170 0.2889573
## 6              frsbe_total  1.63066692 0.1029606

Selection for gift cards

Create a sequence from 1 to 44 numbers <- 1:46

Randomly select 8 numbers selected_numbers <- sample(numbers, 8)

Format the selected numbers with leading zeros formatted_numbers <- sprintf(“%03d”, selected_numbers)

Print the selected numbers print(formatted_numbers)

“020” “007” “006” “019” “024” “041” “013” “037”